Ramachandran's learned paralysis hypothesis: before amputation, the brain repeatedly sent motor commands to a damaged limb and received no response. It learned — correctly — that this hand does not move when commanded. After the amputation, the phantom inherits that learning. The model is what's hurting. Updating the model is what helps.
This simulation traces the mechanism: learning phase, stuck phantom, mirror box therapy. See entry-376.
V. S. Ramachandran proposed the learned paralysis hypothesis in the 1990s to explain why phantom limbs are often stuck in painful positions. Before amputation, many patients had damaged or immobilized limbs. Every motor command went out and returned no proprioceptive confirmation. The brain, confronted with repeated command-without-response, learned: this hand doesn't move when told to.
After amputation, the phantom inherits this learning. The motor system generates commands; the model predicts no movement; and pain persists in a fist that cannot open — in a limb that no longer exists.
The mirror box (Ramachandran & Rogers-Ramachandran, 1996) works by bypassing the motor command loop entirely. The patient places the intact hand in front of the mirror. Looking at the reflection, they see what appears to be both hands. When they move the intact hand, the visual system receives the signal it has been denied: the hand is moving. The brain updates the model. Some patients report the phantom unclenching — and the pain diminishing — within minutes.
What the mirror treats is not tissue. It treats a rule the motor cortex learned — correctly, on good evidence — before the amputation. The rule was never wrong. It just stopped applying. And from inside the model, there was no way to know that.